import numpy as np
import pandas as pd
from sklearn.preprocessing import MinMaxScaler
from sklearn.ensemble import IsolationForest
import tensorflow as tf
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense
import matplotlib.pyplot as plt

i = 1
#canmbiar el valor de csv a agarrar para el entrenamiento
while (i < 71):
    # Path to your CSV
    path = "D:/UsX/Escritorio/space_apps_2024_seismic_detection/data/lunar/training/data/S12_GradeA/" + str(i) + ".csv"
    df = pd.read_csv(path)
    df.rename(columns={'time_abs(%Y-%m-%dT%H:%M:%S.%f)': 'date', 'time_rel(sec)': 'time', 'velocity(m/s)': 'velocity'}, inplace=True)

    # Interpolate missing data
    df.interpolate(method='linear', inplace=True)

    # --- Handle outliers (clip velocity values) ---
    # Clip values that are outside a reasonable range (modify thresholds based on your data)
    df['velocity'] = np.clip(df['velocity'], df['velocity'].quantile(0.01), df['velocity'].quantile(0.99))

    # Normalize the data
    scaler = MinMaxScaler()
    df[['velocity']] = scaler.fit_transform(df[['velocity']])

    # --- Smoothing the signal before feature extraction ---
    df['smoothed_velocity'] = df['velocity'].rolling(window=50, min_periods=1).mean()

    def extract_features(df, window_size=100):
        windows = []
        for start in range(0, len(df), window_size):
            window = df['smoothed_velocity'][start:start+window_size]
            if len(window) == window_size:
                windows.append([
                    np.mean(window),
                    np.std(window),
                    np.max(window),
                    np.min(window),
                    np.median(window)
                ])
        return np.array(windows)

    # Extract features using smoothed velocity
    features = extract_features(df)

    # --- Isolation Forest for anomaly detection ---
    iso_forest = IsolationForest(contamination=0.01, random_state=42)
    iso_forest.fit(features)

    # Predict anomalies (1 = normal, -1 = anomaly)
    anomalies_if = iso_forest.predict(features)

    # Identify anomaly indices
    anomaly_indices_if = np.where(anomalies_if == -1)[0]

    # --- Autoencoder for anomaly detection ---
    autoencoder = Sequential([
        Dense(32, activation='relu', input_shape=(features.shape[1],)),
        Dense(16, activation='relu'),
        Dense(32, activation='relu'),
        Dense(features.shape[1], activation='sigmoid')
    ])

    # Compile and train the autoencoder
    autoencoder.compile(optimizer='adam', loss='mse')
    autoencoder.fit(features, features, epochs=50, batch_size=32, shuffle=True)

    # Reconstruction error
    reconstructions = autoencoder.predict(features)
    mse = np.mean(np.power(features - reconstructions, 2), axis=1)

    # --- Adaptive thresholding based on reconstruction error ---
    threshold = np.percentile(mse, 95)  # Adjust threshold (95th percentile)
    anomalies_ae = mse > threshold
    anomaly_indices_ae = np.where(anomalies_ae)[0]

    # --- Visualization of anomalies ---
    plt.figure(figsize=(10, 6))
    plt.plot(df['time'], df['velocity'], label='Seismic Signal', alpha=0.2)
    
    # Mark anomalies from Isolation Forest in red
    plt.scatter(df['time'][anomaly_indices_if * 100], df['velocity'][anomaly_indices_if * 100], 
                color='red', label='Isolation Forest Anomalies', alpha=0.5)
    
    # Mark anomalies from Autoencoder in green
    plt.scatter(df['time'][anomaly_indices_ae * 100], df['velocity'][anomaly_indices_ae * 100], 
                color='blue', label='Autoencoder Anomalies', alpha=0.5)
    
    plt.xlabel('Time')
    plt.ylabel('Velocity')
    #title ='Anomalies Detected in Seismic Data: ', str(i)
    plt.title('Anomalies Detected in Seismic Data: ' + str(i))
    plt.legend()
    plt.show()

    # Plot the smoothed signal for identifying outliers or spikes
    plt.figure(figsize=(10, 6))
    plt.plot(df['time'], df['smoothed_velocity'])
    plt.xlabel('Time')
    plt.ylabel('Smoothed Velocity')
    plt.title('Smoothed Seismic Signal: ' + str(i))
    plt.show()

    i += 1

C:\Users\UsX\AppData\Local\Temp\ipykernel_22772\2509651175.py:19: FutureWarning: DataFrame.interpolate with object dtype is deprecated and will raise in a future version. Call obj.infer_objects(copy=False) before interpolating instead.
  df.interpolate(method='linear', inplace=True)

Epoch 1/50

c:\Users\UsX\miniconda3\envs\nsa\Lib\site-packages\keras\src\layers\core\dense.py:87: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
  super().__init__(activity_regularizer=activity_regularizer, **kwargs)

179/179 ━━━━━━━━━━━━━━━━━━━━ 1s 955us/step - loss: 0.0317
Epoch 2/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 888us/step - loss: 4.7695e-04
Epoch 3/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 905us/step - loss: 3.3526e-04
Epoch 4/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 905us/step - loss: 2.8795e-04
Epoch 5/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 893us/step - loss: 1.9131e-04
Epoch 6/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 927us/step - loss: 1.5831e-04
Epoch 7/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 860us/step - loss: 1.2092e-04
Epoch 8/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 916us/step - loss: 1.5075e-04
Epoch 9/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 922us/step - loss: 1.9379e-04
Epoch 10/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 933us/step - loss: 8.8691e-05
Epoch 11/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 933us/step - loss: 1.3733e-04
Epoch 12/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 938us/step - loss: 1.0748e-04
Epoch 13/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 882us/step - loss: 1.1896e-04
Epoch 14/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 882us/step - loss: 1.0786e-04
Epoch 15/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 888us/step - loss: 1.0263e-04
Epoch 16/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 899us/step - loss: 1.0752e-04
Epoch 17/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 877us/step - loss: 9.1374e-05
Epoch 18/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 893us/step - loss: 9.2829e-05
Epoch 19/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 871us/step - loss: 9.5435e-05
Epoch 20/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 899us/step - loss: 1.0832e-04
Epoch 21/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 882us/step - loss: 7.2445e-05
Epoch 22/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 882us/step - loss: 9.8456e-05
Epoch 23/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 860us/step - loss: 9.6553e-05
Epoch 24/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 893us/step - loss: 7.2671e-05
Epoch 25/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 899us/step - loss: 5.7939e-05
Epoch 26/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 1ms/step - loss: 7.9480e-05
Epoch 27/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 1ms/step - loss: 7.1042e-05 
Epoch 28/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 1ms/step - loss: 5.5162e-05
Epoch 29/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 961us/step - loss: 5.2110e-05
Epoch 30/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 899us/step - loss: 6.2759e-05
Epoch 31/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 893us/step - loss: 7.3980e-05
Epoch 32/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 922us/step - loss: 5.5900e-05
Epoch 33/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 1ms/step - loss: 4.5367e-05 
Epoch 34/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 1ms/step - loss: 6.4566e-05
Epoch 35/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 1ms/step - loss: 5.2979e-05
Epoch 36/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 995us/step - loss: 4.7184e-05
Epoch 37/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 905us/step - loss: 3.6136e-05
Epoch 38/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 899us/step - loss: 3.5947e-05
Epoch 39/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 927us/step - loss: 2.8142e-05
Epoch 40/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 899us/step - loss: 2.8130e-05
Epoch 41/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 877us/step - loss: 2.2847e-05
Epoch 42/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 905us/step - loss: 2.5675e-05
Epoch 43/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 950us/step - loss: 2.2559e-05
Epoch 44/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 933us/step - loss: 2.0534e-05
Epoch 45/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 882us/step - loss: 1.7190e-05
Epoch 46/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 854us/step - loss: 2.0880e-05
Epoch 47/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 854us/step - loss: 1.7276e-05
Epoch 48/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 983us/step - loss: 2.5583e-05
Epoch 49/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 877us/step - loss: 2.1476e-05
Epoch 50/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 843us/step - loss: 1.7267e-05
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 922us/step

C:\Users\UsX\AppData\Local\Temp\ipykernel_22772\2509651175.py:19: FutureWarning: DataFrame.interpolate with object dtype is deprecated and will raise in a future version. Call obj.infer_objects(copy=False) before interpolating instead.
  df.interpolate(method='linear', inplace=True)

Epoch 1/50

c:\Users\UsX\miniconda3\envs\nsa\Lib\site-packages\keras\src\layers\core\dense.py:87: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
  super().__init__(activity_regularizer=activity_regularizer, **kwargs)

179/179 ━━━━━━━━━━━━━━━━━━━━ 1s 938us/step - loss: 0.0174
Epoch 2/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 893us/step - loss: 7.6279e-04
Epoch 3/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 854us/step - loss: 4.2754e-04
Epoch 4/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 843us/step - loss: 2.5839e-04
Epoch 5/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 899us/step - loss: 2.1537e-04
Epoch 6/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 888us/step - loss: 2.6389e-04
Epoch 7/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 871us/step - loss: 2.0771e-04
Epoch 8/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 843us/step - loss: 2.0860e-04
Epoch 9/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 893us/step - loss: 1.8399e-04
Epoch 10/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 967us/step - loss: 1.8185e-04
Epoch 11/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 995us/step - loss: 1.5185e-04
Epoch 12/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 1ms/step - loss: 1.3606e-04
Epoch 13/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 1ms/step - loss: 1.4439e-04
Epoch 14/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 1ms/step - loss: 1.6140e-04
Epoch 15/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 1ms/step - loss: 1.2067e-04
Epoch 16/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 899us/step - loss: 1.0730e-04
Epoch 17/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 888us/step - loss: 1.0108e-04
Epoch 18/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 837us/step - loss: 1.3080e-04
Epoch 19/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 826us/step - loss: 9.6830e-05
Epoch 20/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 837us/step - loss: 8.0137e-05
Epoch 21/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 865us/step - loss: 5.8792e-05
Epoch 22/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 849us/step - loss: 4.4042e-05
Epoch 23/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 849us/step - loss: 5.1724e-05
Epoch 24/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 843us/step - loss: 3.8520e-05
Epoch 25/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 871us/step - loss: 3.9409e-05
Epoch 26/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 826us/step - loss: 4.6317e-05
Epoch 27/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 888us/step - loss: 3.8968e-05
Epoch 28/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 848us/step - loss: 3.6117e-05
Epoch 29/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 849us/step - loss: 4.3093e-05
Epoch 30/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 826us/step - loss: 3.9186e-05
Epoch 31/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 854us/step - loss: 3.5156e-05
Epoch 32/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 899us/step - loss: 4.0373e-05
Epoch 33/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 820us/step - loss: 3.3426e-05
Epoch 34/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 815us/step - loss: 3.2873e-05
Epoch 35/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 905us/step - loss: 3.5416e-05
Epoch 36/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 860us/step - loss: 3.6700e-05
Epoch 37/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 854us/step - loss: 2.9856e-05
Epoch 38/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 849us/step - loss: 2.8812e-05
Epoch 39/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 893us/step - loss: 3.6373e-05
Epoch 40/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 843us/step - loss: 3.4183e-05
Epoch 41/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 1ms/step - loss: 2.5988e-05
Epoch 42/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 854us/step - loss: 3.0467e-05
Epoch 43/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 944us/step - loss: 2.8169e-05
Epoch 44/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 1ms/step - loss: 2.1376e-05
Epoch 45/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 978us/step - loss: 3.1381e-05
Epoch 46/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 1ms/step - loss: 2.5332e-05
Epoch 47/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 1ms/step - loss: 2.3726e-05
Epoch 48/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 1ms/step - loss: 2.7345e-05
Epoch 49/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 1ms/step - loss: 2.5661e-05
Epoch 50/50
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 2ms/step - loss: 2.7119e-05
179/179 ━━━━━━━━━━━━━━━━━━━━ 0s 1ms/step

C:\Users\UsX\AppData\Local\Temp\ipykernel_22772\2509651175.py:19: FutureWarning: DataFrame.interpolate with object dtype is deprecated and will raise in a future version. Call obj.infer_objects(copy=False) before interpolating instead.
  df.interpolate(method='linear', inplace=True)

Epoch 1/50